Lecture 12: Multiple regression part 2
February 20, 2025 (6th week of classes)Multiple regression (part 2)
This lecture focuses on multiple regression as a fundamental modeling tool, covering both theoretical foundations and practical considerations. It emphasizes the properties of regression models, particularly assumptions about measurement errors in predictors (X) and responses (Y), highlighting how errors in X can distort slope estimates. The lecture also addressed the impact of missing predictors and multicollinearity, demonstrating how correlated missing variables can alter coefficient estimates and model interpretation. It explores the importance of residual variance assumptions (homoscedasticity) and their role in model accuracy and precision. Finally, it introduces key metrics for model selection, hypothesis testing, and diagnostics to evaluate regression models effectively.
Lecture